A domain-independent system for sketch recognition

Abstract

Freehand sketching is a natural and powerful means of interpersonal communication. But to date, it still cannot be supported effectively by human-computer interface. In this paper, we describe a domain-independent system for sketch recognition. Our system allows users to draw sketches as naturally as how they do on paper, and it recognizes the drawing through imprecise stroke approximation which is implemented in a unified and incremental procedure. This method can handle smooth curves and hybrid shapes as gracefully as it does to polylines. With a feature-area verification mechanism and the intelligent adjustment in the post-process, the system can produce user-intended results. Moreover, the output is organized in a hierarchical structure which includes syntactic and semantic information as well as raw data. Our system mainly utilizes low-level geometric features and does not rely on any domain-specific knowledge. Therefore, it will serve as a general and solid foundation for future high-level applications.